class ApplyJaw:
"""
This class will listen to remapped basic_head_api output, apply jaw position
and send the modified message to pau2motors.
"""
rms_params = {"scale": 1.0 / 5000, "min": 0.3, "max": 0.7}
soundfile = None
def hit(self, rms):
"""
Publishes the current jaw-modified expression, given rms (root mean squared),
the volume or the power of a small segment in the file.
"""
# Map the power number to the jaw coefficient.
# Note that coefficient can't effectively go below 0 or over 1.
# It will be cut to this range at the receiving end (pau2motors node)
p = self.rms_params
jaw_coeff = min(max(math.sqrt(rms * p["scale"]), p["min"]), p["max"])
# Copy pau expression message stored during handle_face_in(),
# modify jaw and publish.
cmd = copy.deepcopy(self.facepau)
coeffs = list(cmd.m_coeffs)
coeffs[ShapekeyStore.getIndex("JawOpen")] = jaw_coeff
cmd.m_coeffs = coeffs
self.pub.publish(cmd)
def stop(self):
if self.soundfile:
self.soundfile.stop()
def play(self, filename):
self.stop()
self.soundfile = SoundFile(filename)
self.soundfile.on_playmore = self.hit # Set callback
self.soundfile.play()
def handle_face_in(self, msg):
# Save expression message for later transmission and modification.
self.facepau = msg
# If no sound file is playing, pass the message unmodified
if not self.soundfile or not self.soundfile.is_playing:
self.pub.publish(msg)
def __init__(self):
self.facepau = pau()
#Face expression pau message comes in.
rospy.Subscriber("speechtest_face_in", pau, self.handle_face_in)
#Modified jaw position pau message comes out.
self.pub = rospy.Publisher("cmd_face_pau", pau, queue_size=10)
def play(self, filename):
self.stop()
self.soundfile = SoundFile(filename)
self.soundfile.on_playmore = self.hit # Set callback
self.soundfile.play()
def play(self, filename):
self.stop()
self.soundfile = SoundFile(filename)
self.soundfile.on_playmore = self.hit # Set callback
self.soundfile.play()
class ITFTalker(Thread):
NODE_NAME = 'itf_talker'
pub = rospy.Publisher('itf_next_sentence', String, queue_size=1)
pub_speech_strength = rospy.Publisher('speech_strength', Float32, queue_size=1)
pub_speech_active = rospy.Publisher('speech_active', Bool, queue_size=1)
soundfile = None
rms_params = {"scale": 1.0/5000, "min": 0.0, "max": 1.0}
gletplayer = None
stop_request_received = False
def __init__(self):
Thread.__init__(self)
rospy.init_node(ITFTalker.NODE_NAME, log_level=rospy.INFO)
rospy.Subscriber("itf_talk", String, self.callback)
rospy.Subscriber("itf_talk_stop", String, self.callback_stop)
pyglet.clock._get_sleep_time = pyglet.clock.get_sleep_time
pyglet.clock.get_sleep_time = lambda sleep_idle: pyglet.clock._get_sleep_time(False)
threading.Timer(0.0, pyglet.app.run).start()
rospy.on_shutdown(pyglet.app.exit)
def split_text_rec(self, input_text, max_length=45):
"""
Split a string into substrings which are at most max_length.
"""
combined_text = []
if (len(input_text) > max_length):
while (len(input_text) > max_length):
lastSpace = max_length
section = ""
while (lastSpace > 0):
if (input_text[lastSpace] == ' '):
section = input_text[:lastSpace]
input_text = input_text[lastSpace:]
lastSpace = -1
lastSpace -= 1
combined_text.append(section.strip())
combined_text.append(input_text.strip())
else:
combined_text.append(input_text.strip())
return combined_text
def downloadFile(self, url, fileName):
fp = open(fileName, "wb")
curl = pycurl.Curl()
curl.setopt(pycurl.URL, url)
curl.setopt(pycurl.WRITEDATA, fp)
curl.perform()
curl.close()
fp.close()
def getGoogleSpeechURL(self, phrase):
googleTranslateURL = "http://translate.google.com/translate_tts?tl=en&"
parameters = {'q': phrase}
data = urllib.urlencode(parameters)
googleTranslateURL = "%s%s" % (googleTranslateURL,data)
return googleTranslateURL
def speakSpeechFromText(self, phrase):
phraseSections = self.split_text_rec(phrase)
for index, section in enumerate(phraseSections):
print "At index " + str(index) + " is " + section + "\n"
for index, section in enumerate(phraseSections):
googleSpeechURL = self.getGoogleSpeechURL(section)
print "Downloading " + googleSpeechURL + " to " + "tts" + str(index).zfill(index) + ".mp3\n"
self.downloadFile(googleSpeechURL,"tts" + str(index).zfill(index) + ".mp3")
print index, section
totalDuration = 0
for index, section in enumerate(phraseSections):
fileName = 'tts' + str(index).zfill(index) + '.mp3'
dubsegment = pydub.AudioSegment.from_mp3(fileName)
totalDuration += dubsegment.__len__()
dubsegment = None
# for index, section in enumerate(phraseSections):
# print 'Calling mplayer with parameter ' + 'tts' + str(index).zfill(index) + '.mp3'
# subprocess.call(['mplayer', 'tts' + str(index).zfill(index) + '.mp3'])
# ITFTalker.pub.publish("Google Voice completed.")
# for index, section in enumerate(phraseSections):
# fileName = 'tts' + str(index).zfill(index) + '.mp3'
# gletsource = pyglet.media.load(fileName, streaming=False)
# self.gletplayer = pyglet.media.Player()
# self.gletplayer.queue(gletsource)
# self.gletplayer.play()
ITFTalker.pub_speech_active.publish(True)
for index, section in enumerate(phraseSections):
if (not self.stop_request_received):
fileName = 'tts' + str(index).zfill(index) + '.mp3'
print 'Calling SoundPlayer with parameter ' + fileName
self.play(fileName)
if not (self.soundfile is None):
while self.soundfile.is_playing:
pass
else:
pass
ITFTalker.pub.publish("Google Voice completed.")
self.pub_speech_strength.publish(0)
ITFTalker.pub_speech_active.publish(False)
self.stop_request_received = False
#os.system("mplayer tts " + str(index).zfill(index) + ".mp3 -af extrastereo=0 &")
def hit(self, rms):
"""
Publishes the current jaw-modified expression, given rms (root mean squared),
the volume or the power of a small segment in the file.
"""
# Map the power number to the jaw coefficient.
# Note that coefficient can't effectively go below 0 or over 1.
# It will be cut to this range at the receiving end (pau2motors node)
p = self.rms_params
jaw_coeff = min(max(math.sqrt(rms * p["scale"]), p["min"]), p["max"])
self.pub_speech_strength.publish(jaw_coeff)
# Copy pau expression message stored during handle_face_in(),
# modify jaw and publish.
# cmd = copy.deepcopy(self.facepau)
# coeffs = list(cmd.m_coeffs)
# coeffs[ShapekeyStore.getIndex("JawOpen")] = jaw_coeff
# cmd.m_coeffs = coeffs
# self.pub.publish(cmd)
def stop(self):
if self.soundfile:
self.soundfile.stop()
def play(self, filename):
self.stop()
self.soundfile = SoundFile(filename)
self.soundfile.on_playmore = self.hit # Set callback
self.soundfile.play()
def callback(self, data):
rospy.loginfo(rospy.get_caller_id()+"I heard %s",data.data)
self.speakSpeechFromText(data.data)
def callback_stop(self, data):
self.stop_request_received = True
self.stop()
from SoundFile import SoundFile
from Sequencer import Sequencer
if __name__ == "__main__":
x = 1
o = 0
sf1 = SoundFile("../sound/drums/1-kick/kick-allaboutyou-1.wav")
sq1 = Sequencer(sf1, [x, o, o, o, x, o, o, o])
sq1.play(5)
sf2 = SoundFile("../sound/drums/4-hihat/closedhh-allaboutyou.wav")
sq2 = Sequencer(sf2, [x, o, x, o, x, o, x, o])
sq2.play(5)
sf3 = SoundFile("../sound/drums/3-snare/snare-allaboutyou.wav")
sq3 = Sequencer(sf3, [o, o, o, o, x, o, o, o])
sq3.play(5)
from SoundFile import SoundFile
from Sequencer import Sequencer
if __name__ == "__main__":
# the path to your sound file
sound = SoundFile("hello.wav")
# sequencer takes a SoundFile as an argument; a list with the sequence; and the tempo in beats per minute (default 120)
seq = Sequencer(sound,[1])
# number of times to play. 0 = infinite. CTRL+C to quit program.
seq.play(1)
class ITFTalker(Thread):
NODE_NAME = 'itf_talker'
#pub = rospy.Publisher('itf_next_sentence', String, queue_size=1)
pub_speech_strength = rospy.Publisher('/act/tts/get_speech_strength',
Float32,
queue_size=1)
pub_speech_active = rospy.Publisher('/act/tts/get_speech_active',
Bool,
queue_size=1)
soundfile = None
rms_params = {"scale": 1.0 / 5000, "min": 0.0, "max": 1.0}
gletplayer = None
stop_request_received = False
def __init__(self):
Thread.__init__(self)
rospy.init_node(ITFTalker.NODE_NAME, log_level=rospy.INFO)
#rospy.Subscriber("itf_talk", String, self.callback)
#rospy.Subscriber("itf_talk_stop", String, self.callback_stop)
rospy.Subscriber("/act/tts/set_text", String, self.callback)
rospy.Subscriber("/act/tts/stop", String, self.callback_stop)
pyglet.clock._get_sleep_time = pyglet.clock.get_sleep_time
pyglet.clock.get_sleep_time = lambda sleep_idle: pyglet.clock._get_sleep_time(
False)
threading.Timer(0.0, pyglet.app.run).start()
rospy.on_shutdown(pyglet.app.exit)
def split_text_rec(self, input_text, max_length=95):
"""
Split a string into substrings which are at most max_length.
"""
combined_text = []
if (len(input_text) > max_length):
while (len(input_text) > max_length):
lastSpace = max_length
section = ""
while (lastSpace > 0):
if (input_text[lastSpace] == ' '):
section = input_text[:lastSpace]
input_text = input_text[lastSpace:]
lastSpace = -1
lastSpace -= 1
combined_text.append(section.strip())
combined_text.append(input_text.strip())
else:
combined_text.append(input_text.strip())
return combined_text
def downloadFile(self, url, fileName):
fp = open(fileName, "wb")
curl = pycurl.Curl()
curl.setopt(pycurl.URL, url)
curl.setopt(pycurl.WRITEDATA, fp)
curl.perform()
curl.close()
fp.close()
def getGoogleSpeechURL(self, phrase):
googleTranslateURL = "http://translate.google.com/translate_tts?tl=en&"
parameters = {'q': phrase}
data = urllib.urlencode(parameters)
googleTranslateURL = "%s%s" % (googleTranslateURL, data)
return googleTranslateURL
def speakSpeechFromText(self, phrase):
phraseSections = self.split_text_rec(phrase)
for index, section in enumerate(phraseSections):
print "At index " + str(index) + " is " + section + "\n"
for index, section in enumerate(phraseSections):
googleSpeechURL = self.getGoogleSpeechURL(section)
print "Downloading " + googleSpeechURL + " to " + "tts" + str(
index).zfill(index) + ".mp3\n"
self.downloadFile(googleSpeechURL,
"tts" + str(index).zfill(index) + ".mp3")
#mandeep - download to mounted ramdisk
print index, section
totalDuration = 0
for index, section in enumerate(phraseSections):
fileName = 'tts' + str(index).zfill(index) + '.mp3'
dubsegment = pydub.AudioSegment.from_mp3(fileName)
totalDuration += dubsegment.__len__()
dubsegment = None
# for index, section in enumerate(phraseSections):
# print 'Calling mplayer with parameter ' + 'tts' + str(index).zfill(index) + '.mp3'
# subprocess.call(['mplayer', 'tts' + str(index).zfill(index) + '.mp3'])
# ITFTalker.pub.publish("Google Voice completed.")
# for index, section in enumerate(phraseSections):
# fileName = 'tts' + str(index).zfill(index) + '.mp3'
# gletsource = pyglet.media.load(fileName, streaming=False)
# self.gletplayer = pyglet.media.Player()
# self.gletplayer.queue(gletsource)
# self.gletplayer.play()
ITFTalker.pub_speech_active.publish(True)
for index, section in enumerate(phraseSections):
if (not self.stop_request_received):
fileName = 'tts' + str(index).zfill(index) + '.mp3'
print 'Calling SoundPlayer with parameter ' + fileName
self.play(fileName)
if not (self.soundfile is None):
while self.soundfile.is_playing:
pass
else:
pass
#ITFTalker.pub.publish("Google Voice completed.")
self.pub_speech_strength.publish(0)
ITFTalker.pub_speech_active.publish(False)
self.stop_request_received = False
#os.system("mplayer tts " + str(index).zfill(index) + ".mp3 -af extrastereo=0 &")
def hit(self, rms):
"""
Publishes the current jaw-modified expression, given rms (root mean squared),
the volume or the power of a small segment in the file.
"""
# Map the power number to the jaw coefficient.
# Note that coefficient can't effectively go below 0 or over 1.
# It will be cut to this range at the receiving end (pau2motors node)
p = self.rms_params
jaw_coeff = min(max(math.sqrt(rms * p["scale"]), p["min"]), p["max"])
self.pub_speech_strength.publish(jaw_coeff)
# Copy pau expression message stored during handle_face_in(),
# modify jaw and publish.
# cmd = copy.deepcopy(self.facepau)
# coeffs = list(cmd.m_coeffs)
# coeffs[ShapekeyStore.getIndex("JawOpen")] = jaw_coeff
# cmd.m_coeffs = coeffs
# self.pub.publish(cmd)
def stop(self):
if self.soundfile:
self.soundfile.stop()
def play(self, filename):
self.stop()
self.soundfile = SoundFile(filename)
self.soundfile.on_playmore = self.hit # Set callback
self.soundfile.play()
def callback(self, data):
rospy.loginfo(rospy.get_caller_id() + "I heard %s", data.data)
self.speakSpeechFromText(data.data)
def callback_stop(self, data):
self.stop_request_received = True
self.stop()
from SoundFile import SoundFile
from Sequencer import Sequencer
if __name__ == "__main__":
# in simplified music notation, usually x = sound, o = no sound
x = 1
o = 0
# the path to your sound file
sf1 = SoundFile("../sound/drums/1-kick/kick-allaboutyou-1.wav")
# sequencer takes a SoundFile as an argument; a list with the sequence; and the tempo in beats per minute (default 120)
sq1 = Sequencer(sf1,[x,o,o,o,], 120)
# number of times to play. 0 = infinite
sq1.play(5)
# have fun!
